CGAL users discussion list

Quoting "Ben Supnik" 
<>:

> Hi Efi,
>
> > It requires exact constructions when it constructs new geometric  
> > object, for example, intersections points. If, for example, the  
> > sweep line is used to construct an arrangement of segments that do  
> > not intersect each other in their interior, then exact  
> > constructions is not required. Exact predicates are always required.
>
> So if I understand correctly, the issue here is that internally  
> sweep line always constructs intersections (when they exist) even if  
> the visitor that is layered on top of it doesn't actually use them  
> (e.g. "do_intersect", as opposed to "return the intersections")?

Looking back at this (short) thread I realize that the question was  
about do_intersect. I believe that exact predicates are not required,  
as new geometric objects should not be generated.

I also notice that the coordinates of the input point in Konstantin  
original message are given in pairs of floating points. This could be  
the source of the problem. Try to replace the floats with strings (or  
integers if they can be represented using 32 bits).

> Out of curiosity, is it theoretically possible to code sweep-line  
> only in terms of predicates on the source curve set (assuming the  
> output format didn't need to return the intersections)?

The sweep line code in implemented in terms of predicates. The only  
constructions are performed by the geometry traits-class operations,  
and those are applied only when needed.


> cheers
> ben
> --
> You are currently subscribed to cgal-discuss.
> To unsubscribe or access the archives, go to
> https://lists-sop.inria.fr/wws/info/cgal-discuss